Pre-coding is a kind of beam forming that supports multi-layer transmission in MIMO radio systems. Simulations have shown that pre-coding increases the performance of wireless communication systems and it has therefore been included in the standardization of LTE (Long Term Evolution). In pre-coding in LTE, each of the multiple streams are emitted from the transmit antennas at the radio base station (RBS) with independent phase and equal amplitude per each antenna such that the throughput is optimized between the radio base station and the user equipment (UE) e.g. mobile. The pre-coding weights are typically calculated at the UE and then the UE informs the radio base station which pre-coding weights that should be used.
Pre-coding can be frequency selective or wideband. In frequency selective pre-coding, PMIs (Precoding Matrix Indicators) can be chosen differently in different frequency sub-bands while a single CQI (Channel Quality Indicator) is chosen for the whole frequency band. In wideband pre-coding, the same PMI is used over the entire frequency band but CQI can be chosen differently in different frequency sub-bands. CQI is a representation of the downlink channel conditions for the user equipment, whereas the PMI represents the recommended pre coder matrix corresponding to the rank e.g. number of independent data streams, depending on estimates of the downlink channel conditions.
One way to increase the channel capacity for a MIMO channel is to use so called water filling. Water filling includes allocating different amount of power to different data streams depending on the eigenvalues of the channel. Simulation results have shown that it is possible to gain from 20-50% in channel capacity when using water filling instead of uniform power allocation for 4×4 MIMO during ideal assumptions and for SINR between 0-20 dB. The gains will probably be even greater for higher order MIMO.
One difficulty with for example LTE is that even though different streams have different SINR for different frequencies, the same power is typically allocated to each stream over the whole frequency band. This is not optimal in the sense of maximizing the throughput. Therefore, there is a need to find methods of improving the power allocation to optimize the throughput.